Estimation of Monotone Treatment Effects in Network Experiments

• Activity Number: 525
• Type: Topic Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): David Choi*
• Companies: Carnegie Mellon University
• Keywords: causal inference ; network data ; interference ; experiments ; social networks

Abstract

Randomized experiments in network settings pose statistical challenges due to the possibility of interference between units. We propose a new method for estimating attributable treatment effects under interference. The method does not require partial interference to hold, but instead uses an identifying assumption that is similar to requiring nonnegative treatment effects. Observed pre-treatment social network information can be used to customize the test statistic, which may increase power and does not add assumptions on the data generating process. The inversion of the test statistic is a combinatorial optimization problem which has a tractable relaxation, yielding conservative one-sided estimates of the attributable effect.